With AMD and NVIDIA already established as the two (and only) big players in the GPU industry, it’s fascinating to see how Intel is trying to break through this duopoly. It’s not like the Arc A-Series is the company’s first-ever foray into the world of discrete graphics either. Intel has plenty of experience tuning graphics on APUs and have even dabbled in desktop releases before, albeit very limitedly.
However, Arc does mark the first time the Blue Team has taken this as seriously as AMD and NVIDIA in order to steal some of that market share. So far, we’ve seen only one desktop release in the form of the Arc A380 GPU, while Intel has launched a few mobile SKUs already.
Due to their limited, or some may say hastily-executed, launch these GPUs are only available in certain parts of the world, mostly Asia. That’s why large-scale testing and evaluation of Arc SKUs peer-reviewed by multiple sources has simply been impossible so far. We have rough ideas pertaining to how each Arc GPU performs, but nothing concrete enough to draw conclusions.
Intel has taken an interesting route to counteract all of this. The company, despite being in graphical infancy, has released official benchmarks for some of their models themselves. And the transparency at display here is remarkable. For instance, yesterday Intel shared benchmark results comparing its Arc A380 GPUs against its AMD and NVIDIA contemporaries, in a test which clearly shows that the GTX 1650 is still faster.
The whole point of releasing benchmarks yourselves is that you can play around with the footnotes to bend the results in your favor, but to see Intel not engage in any malpractice (to some extent), especially given how crucial this launch is for the company, is, once again, quite commendable.
Arc A770M and A730M official benchmark
Now, just a day after the Arc A380 benchmarks came out, Intel has gone ahead and released the official gaming benchmarks for two of its higher-end mobile GPUs: Arc A730M and Arc 770M. This data was provided with the Reviewer’s Guide but that was only made available to Chinese reviewers.
Tom’s Hardware somehow managed to get their hands on this data for the whole world to see. Thought, keep in mind we can’t validate this claim since Tom’s Hardware did not provide an official document, rather just the numbers. Nonetheless, the results are still pretty interesting.
Interestingly, both GPUs were compared to lower-tier chips than expected. The A770M went up against the RTX 3060 mobile whereas the Arc A730M went up against the RTX 3050 Ti mobile. Previously, it was expected that the Arc A770M would be on-par with the RTX 3070 instead, but it seems like Intel has backtracked from their estimates. For a while, RTX 3080 series was also being considered as a possible contented for the top-end Arc mobile SKUs, but clearly that is no longer the case.
Before we look at the benchmarks, it’s important to mention that this test isn’t exactly linear and definitely not as transparent as yesterday’s.
For one, Intel did not provide exact TGP targets for either of these GPUs, that way we can’t know what TGP they’re actually running at during these tests. Similarly, Intel did not use the fully-powered versions of the RTX 3060 and 3050 Ti mobile, whereas Arc A770M and A730M more than likely were running at their maximum target TGP. But even that wasn’t confirmed by Intel.
GPU & test bench specifications
Moreover, the ‘test bench’ used for each GPU varied wildly, as well. For the RTX 3050 Ti, Intel used the ROG Zephyrus M16 with an 11th Gen Core i5 processor and 16GB of DDR4 RAM. For the RTX 3060 mobile, the company used the MSI Pulse GL66 featuring the flagship Alder Lake mobile CPU with DDR5 RAM instead. For its own GPUs, a Core i7 was paired with the A730M but an i9 was paired with the A770M.
So, the laptops used here act as another variable that could affect the performance for any of these GPUs. Following are the details we know about the configurations:
- Intel Arc A730M – 80W to 120W: Core i7-12700H, 16GB of DDR5-4800
- Intel Arc A770M – 120W to 150W: Core i9-12900HK, 16GB of DDR5-4800
- NVIDIA GeForce RTX 3050 Ti mobile – 60W: Core i7-11800H, 16GB of DDR4-3200 [ROG Zephyrus M16]
- NVIDIA GeForce RTX 3060 mobile – 85W: Core i9-12900HK, DDR5-4800 [MSI Pulse GL66]
As for the actual specifications of the Arc SKUs, thankfully Intel did still provide them but we don’t have any official document for that either. Regardless, these are still official.
The Arc A730M and A770M both use the ACM-G10 GPU and are part of the Arc 7 series (duh). The Arc A770M utilizes the full ACM-G10 GPU and therefore features 32 Xe-Cores and 32 Ray Tracing units. It has a base clock speed of 1650MHz, up to 16GB of GDDR6 memory running across a 256-bit wide bus with a target TDP of 120-150W. This is Intel’s best mobile GPU to date.
On the other hand, the Arc A730M is no slouch either. It uses a cut-down version of the same ACM-G10 GPU so it only gets 24 Xe Cores and, thus, 24 Ray Tracing units. Moreover, it features 3072 ALUs, an 1100MHz base clock speed, 12GB of GDDR6 memory operating across a 192-bit bus interface, with a TDP target set around 80-120W.
Benchmark results (gaming)
Now that we’re up to mark with the specs of the SKUs, let’s take a look at the benchmark results comparing the Arc A770M to the RTX 3060 and the Arc A730M to the RTX 3050 Ti:
Taking a look at the results, we can see that both Arc GPUs are faster than what they’re compared to, overall. The A770M is around 13% faster on average than the RTX 3050 Ti, whereas the A770M beats the RTX 3060 Ti with 12% better frames all around.
Out of the total 17 games tested, the Arc GPU was better in nearly every single one of them compared to its competition. Keep in mind that these games were tested at the highest settings possible, so Ultra and High where applicable. But the resolution was kept at only 1080p. Perhaps, a more GPU bound test at 1440p resolution can provide even clearer results, but we’ll leave that aside for now.
As for whether you should take these numbers at face value or not, well, that’s entirely up to you. Intel’s Arc A380 benchmarks showed that their GPU was strong and capable, but not quite ready yet, and Intel showed all that themselves! So, maybe a similar mentality is being adopted here. After all, the Arc GPUs do lose in a few games in this test.
The biggest takeaway from this benchmark is that Intel Arc GPUs’ actual TGP is unknown while the NVIDIA GPUs are running on much less power than they can. The RTX 3060 mobile used in this test, for instance, is running at 85W, however it can go up to 115W to offer better performance. Meanwhile, Arc A770M could be running anywhere between 120 and 150W, yet that would still make it much less efficient than NVIDIA’s offerings.
Not enough evidence
Intel needs to send these GPUs to as many publications as possible to fact-check their claims and compare the date against their own evaluations to draw true conclusive answers. In fact, Arc A770M hasn’t even been released yet but we’re seeing its gaming performance alongside the Arc A730M, a GPU that is currently available in China.
Since these are mobile GPUs, we have no idea about pricing. An educated guess would be to expect these to power high-end ultrabooks or thin-and-light gaming laptops in the future. But that’s the thing; we don’t know what the future holds yet. With AMD and NVIDIA right on Intel’s tail with their next generation of graphics cards, it doesn’t look like the Blue Team can hold on for much longer.
Arc GPUs are still not available worldwide and coordinating a global launch will be the true Mount Everest for Intel to summit. If the company does this right and people actually get their hands on Arc before RDNA 3 or RTX 40-series is here, Intel might just have a chance at carving out a niche for itself. But, then again, scalpers await.